Email Marketing MCP for DTC Brands
DTC brands live and die by inbox performance. marginal gives your AI agent a direct line to run subject line A/B tests, launch managed sends, and read back lift metrics — all over a single hosted email marketing MCP endpoint.
Why DTC brands wire email into an MCP server
Direct-to-consumer storefronts ship campaigns constantly: product drops, restock alerts, win-back flows, and seasonal promos. The bottleneck is rarely creative — it's the loop of writing variants, launching a test, and waiting on numbers before the next send.
An email MCP server collapses that loop. Instead of bouncing between a dashboard and a spreadsheet, you ask your agent to test and report, and the tools do the work behind a consistent interface.
- Test launch-day subject lines without leaving your editor
- Keep experiment history close to the code that generates content
- Let an AI agent handle the repetitive send-and-measure cycle
- Start free with 100 experiments/month — enough for most DTC cadences
The tools behind your e-commerce email experiments
marginal exposes four MCP tools that map cleanly to how DTC teams already think about email. Your agent calls them in sequence to run a complete subject line A/B test.
- generate_variants — produce subject line options tuned to a launch or promo
- launch_test — start a managed send and split your list across variants
- get_results — pull open/click tracking and within-test lift metrics
- recommend_next — get a data-backed suggestion for the follow-up send
A typical DTC test loop with marginal
Say you're prepping a Friday product drop. You prompt your agent to draft five subject lines for the announcement, launch a split test to a sample of your list, and check results before the full send goes out.
Because the AI agent email marketing flow runs through one MCP endpoint, the same conversation that writes copy also reads back which variant won. That tight feedback makes e-commerce email experiments cheap enough to run on every meaningful campaign, not just the big ones.
- Drop announcements: test urgency vs. curiosity framing
- Restock alerts: compare product-name-first vs. benefit-first lines
- Win-back: A/B test discount-led vs. value-led subjects
- Cart recovery: measure lift before scaling a winning line
Connect your agent to marginal
marginal is hosted, so there's no MCP server to deploy or maintain. Point a supported client at the endpoint, authenticate with your Bearer API key, and the tools appear in your agent.
Works with Cursor, Claude Desktop, Claude Code, OpenAI Codex, Windsurf, Cline, Continue, and Zed. The endpoint is https://marginal.sh/mcp, and full setup steps live in the docs at https://marginal.sh/docs/.
- Endpoint: https://marginal.sh/mcp
- Registry: sh.marginal/mcp
- Auth: Bearer API key (marg_live_...)
- No self-hosting — managed sends and tracking run on marginal
Get started with marginal
marginal is a hosted email marketing MCP server at marginal.sh. Sign up free, create an API key, and connect https://marginal.sh/mcp from Cursor, Codex, or Claude Desktop.
- 100 experiments/month on the free tier
- Four MCP tools: generate_variants, launch_test, get_results, recommend_next
- Listed in the MCP Registry as sh.marginal/mcp